Destabilization of
␣
-Helical Structure in Solution Improves
Bactericidal Activity of Antimicrobial Peptides: Opposite Effects on
Bacterial and Viral Targets
David O. Ulaeto,aChristopher J. Morris,bMarc A. Fox,aMark Gumbleton,cKonrad Beckd CBR Division, Dstl Porton Down, Salisbury, United Kingdoma
; School of Pharmacy, University of East Anglia, Norwich, United Kingdomb
; College of Biomedical and Life Sciences, School of Pharmacy and Pharmaceutical Sciences, Cardiff University, Cardiff, United Kingdomc
; College of Biomedical and Life Sciences, School of Dentistry, Cardiff University, Cardiff, United Kingdomd
We have previously examined the mechanism of antimicrobial peptides on the outer membrane of vaccinia virus. We show here that the formulation of peptides LL37 and magainin-2B amide in polysorbate 20 (Tween 20) results in greater reductions in virus titer than formulation without detergent, and the effect is replicated by substitution of polysorbate 20 with high-ionic-strength buffer. In contrast, formulation with polysorbate 20 or high-ionic-strength buffer has the opposite effect on bactericidal activity of both peptides, resulting in lesser reductions in titer for both Gram-positive and Gram-negative bacteria. Circular dichroism spectroscopy shows that the differential action of polysorbate 20 and salt on the virucidal and bactericidal activities correlates
with the␣-helical content of peptide secondary structure in solution, suggesting that the virucidal and bactericidal activities are
mediated through distinct mechanisms. The correlation of a defined structural feature with differential activity against a host-derived viral membrane and the membranes of both Gram-positive and Gram-negative bacteria suggests that the overall helical content in solution under physiological conditions is an important feature for consideration in the design and development of candidate peptide-based antimicrobial compounds.
A
ntimicrobial peptides (AMPs) exhibit potent, broad-spec-trum bactericidal activity and have been shown to be active against a limited number of viruses (1). They are typically 15 to 40 amino acids in length and variously display the structural charac-teristics of␣-helices,-sheets, or disordered chains. Some AMPs combine multiple structural elements, and many are stabilized or constrained by intramolecular disulfide bonds or cyclized into circular or loop structures (reviewed in reference1).Helical content is an important factor in the microbicidal ac-tivity of AMPs, with increased helical content in hydrophobic en-vironments or lipid or detergent micelles seen as a marker for potent microbicidal activity (2–5). For many linear peptides, physiological salt concentrations inhibit microbicidal activity (6, 7). However,␣-helical structure of cathelicidin peptides is pro-moted by elevating the ionic strength of buffers, which correlates with inhibition of antimicrobial activity (8,9), and there is con-tinuing debate on the mechanistic function of secondary structure for antimicrobial activity. The profound microbicidal activity of AMPs warrants a fuller understanding of their mechanisms of action in order to realize their potential as treatments that avoid resistance mechanisms that apply to conventional antibiotics.
LL37 is a 37-mer␣-helical AMP derived posttranslationally from human cathelicidin. It has microbicidal activity against both Gram-positive and Gram-negative bacteria and viruses (10). Studies with multiple primate cathelicidins show that LL37 homo-logues with lower propensity to adopt␣-helical conformation in solution have more potent bactericidal activity, which is less sus-ceptible to inhibition by physiological concentrations of salt (8,9). LL37 itself will readily adopt␣-helical conformation in solution; this is enhanced at physiological salt concentrations, while its bac-tericidal activity is inhibited.
Salt-dependent conformation change is also a feature of the
Xenopus laevisAMP magainin, which adopts a greater degree of
␣-helicity in solutions containing increased salt, while its bacteri-cidal activity is reduced (3,11–14). Although magainin has mainly a disordered conformation at physiological salt concentrations, it has been shown to adopt an␣-helical conformation in lipid mem-branes (2,3,15). Enhanced␣-helical folding in lipid membranes is also observed for LL37 (5), and it is thus possible that whereas adoption of␣-helical structure is important for antimicrobial ac-tivity of this class of AMP, the point at which the AMP adopts the
␣-helical structure is equally important, and that preformed heli-ces are less able to disrupt the target membranes.
We have previously shown that LL37 and alanine substituted (S8A, G13A, and G18A) magainin-2 amide (magainin-2B) could inactivate vaccinia virus (VACV) but were unable to inactivate all the virus in a preparation, generally leaving a viable rump of ca. 10% of the starting preparation (16). In this study, we sought to improve the activity of AMPs against VACV by formulation with the nonionic detergent, polysorbate 20 (Tween 20). Experiments were also undertaken with Gram-positiveStaphylococcus aureus
and Gram-negativeEscherichia colibacteria to determine whether
Received3 September 2015Returned for modification24 October 2015 Accepted30 December 2015
Accepted manuscript posted online11 January 2016
CitationUlaeto DO, Morris CJ, Fox MA, Gumbleton M, Beck K. 2016.
Destabilization of␣-helical structure in solution improves bactericidal activity of antimicrobial peptides: opposite effects on bacterial and viral targets. Antimicrob Agents Chemother 60:1984 –1991.doi:10.1128/AAC.02146-15.
Address correspondence to David O. Ulaeto, [email protected].
Supplemental material for this article may be found athttp://dx.doi.org/10.1128 /AAC.02146-15.
© Crown copyright 2016. This is an open-access article distributed under the terms of theCreative Commons Attribution 4.0 International license.
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
modulation of microbicidal activity applied equally to different types of microbial membranes. We analyzed the effect of the de-tergent on the secondary structure of LL37 and magainin-2B by circular dichroism (CD) spectroscopy and correlated this with the effect of ionic strength on secondary structure and with the effect of both detergent and ionic strength on activity againstS. aureus,
E. coli, and VACV.
MATERIALS AND METHODS
Reagents.Unless otherwise stated, all reagents were purchased from Sigma-Aldrich Chemical Company.
Cell lines and viruses.Enveloped virions of the IHD-J strain of vac-cinia virus were harvested at 18 to 20 h postinfection from the culture supernatant of RK13 rabbit epithelial cells synchronously infected at a multiplicity of infection ofⱖ10 to provide virions with undamaged enve-lopes. All cell and virus cultures were maintained in DMEM with 10% fetal bovine serum, 2 mML-glutamine, and 100 U of penicillin and strep-tomycin ml⫺1(tissue culture medium) at 37°C in a 5% CO2, humidified
atmosphere. Virus was used fresh on the day of preparation and was quantitated by limiting dilution titration on RK13 cells, and the 50% tissue culture infectious dose was calculated by Reed-Muench analysis of virus-positive wells (17,18). The entire volume of the sample was used in titrations. The requirement for virus to be used as a fresh preparation meant that starting virus titer could not be determined in advance of the experiments and was always determined from control samples as part of the experiment.
Bacterial strains.E. colistrain XL1-Blue was obtained from Strat-agene, andS. aureusstrain ATCC-29213 was obtained from the American Type Culture Collection. For bacterial cultures, a single colony grown on L agar plates (S. aureus) or a loop from a frozen ampoule obtained direct from the supplier (E. coli) was inoculated directly into 5 ml of L broth in a vented T25 tissue culture flask. Two serial dilutions were taken using fresh loops into further T25 flasks containing 5 ml of L broth, and all three flasks were cultured overnight, upright in a shaking incubator at 37°C, after which the most dilute flask with visual turbidity was selected. A sample of the culture was diluted in H2O to give an absorbance reading of between
0.01 and 0.02 at 600 nm, and this was used for incubation with AMPs. After incubation with AMPs for 10 min, bacterial cultures were serially diluted in steps of 1:3, applied in spots to L agar plates (8 spots per dilu-tion, in 10-l volumes), and cultured overnight at 37°C. After this, each spot was evaluated for bacterial growth, and the data were used to calcu-late a 50% colony-forming dose by the method of Reed and Muench (17, 18). The entire volume of the sample was used in titrations.
Peptides.Peptides were synthesized by Alta Biosciences, Birmingham, United Kingdom. The LL37 sequence used was LLGDFFRKSKEKIGKEF KRIVQRIKDFLRNLVPRTES (19). The magainin-2B sequence used was GIGKFLHAAKKFAKAFVAEIMNS, with the C terminus amidated (11). Unless otherwise stated, peptide stocks were made at 1 mg/ml in phos-phate-buffered saline (PBS).
Peptide treatments.Peptides were dispensed into tubes at a 10⫻ con-centration. Negative controls received an equivalent volume of PBS with no peptide. For treatment in the presence of polysorbate 20 (Tween 20), an appropriate volume of 10⫻concentration (vol/vol in H2O)
polysor-bate 20 or H2O was added to the peptide and control samples and then
mixed by vortexing. Finally, the virus or bacteria were added, and the samples were vortexed and incubated at 37°C. Sample volumes were al-ways 50l. At the end of the incubation, samples were diluted by the addition of 130l of tissue culture medium for virus or 130l of L broth for bacteria and then quantitated as described above. In each microbicidal activity experiment, all samples were prepared in triplicate, and all data are presented as means and standard deviations of triplicates.
CD spectroscopy.CD spectra were recorded on an Aviv model 215 spectropolarimeter (Aviv Biomedical, Inc., Lakewood, NJ) equipped with a thermostated cell holder. Far-UV spectra (260 toⱕ190 nm) were col-lected in a 0.1-cm quartz cuvette at peptide concentrations of 0.1 to 0.2
mg/ml determined spectroscopically (20). Buffer baselines (with or with-out polysorbate 20) recorded in the same cell were subtracted, and data were smoothed and normalized to mean residue ellipticities ([⍜]MRW).
The instrument was calibrated with camphorsulfonic acid (21). Lyophi-lized peptides were dissolved in 150 mM NaF and 20 mM KH2PO4/NaOH
(pH 7.4). NaF was used instead of NaCl due to the strong absorbance of chloride ions belowⱕ195 nm. Polysorbate 20 in the same buffer was added to peptide samples at least 1 h prior to measurements.
The fraction of␣-helical peptide conformation was calculated as de-scribed previously (22). Detergent-induced␣-helix formation was mon-itored at 222 nm, and data were fitted assuming a one-site binding mech-anism, fH⫽f
hmax⫻c/(Kd⫹c), where fHis the fraction of helix, fhmaxis
the maximum helix fraction achievable, andKdis the apparent
dissocia-tion constant.
Statistical analysis.For statistical analysis, the raw data were trans-formed by taking the log10of each data-point prior to analysis of variance
(ANOVA). The experimental design facilitates the combination of repeat experiments into a single analysis using a two-way ANOVA where one of the parameters is variation between experiments. It is to be expected that in some cases there will be significant variation between repeat experi-ments because the starting titer of virus and bacterial preparations is not known in advance. However, in all cases where there was a significant variation between experiments, the analysis showed that there was no interaction between the interexperiment variation and the effects, if any, of AMP treatment.
RESULTS
Virucidal activity of AMPs is enhanced synergistically by
poly-sorbate 20.We have previously shown that LL37 has greater
viru-cidal activity against VACV than magainin-2B but that both re-move the outer membrane of the virus (16). The efficacy of LL37 in inactivating VACV led us to examine methods by which the activity of magainin-2B might be improved. Thus, we took prep-arations of VACV and treated them with LL37 or magainin-2B in the presence or absence of the nonionic detergent, polysorbate 20, at a concentration above the critical micellar concentration, on the assumption that the detergent would alter the secondary struc-ture of the AMPs. LL37 treatment significantly reduced the quan-tity of virus recovered (P⫽0.004) while magainin-2B had a lesser and insignificant effect (P⫽0.1) in the absence of polysorbate 20 (Fig. 1A). When incubated with virus in the presence of polysor-bate 20 for 1 h, magainin-2B acquired significant virucidal activity (⬍0.001) and a significant enhancement of the activity of LL37 (P⫽0.004) was also observed (Fig. 1B). Treatment with polysor-bate 20 alone for 60 min produced a slight increase in titer, pre-sumably as a result of disaggregation of virus particles (data not shown).
The improved activity of both LL37 and magainin-2B against VACV in the presence of polysorbate 20 led us to determine whether the AMP and the detergent were acting independently or cooperatively. Virus was treated with magainin-2B and subse-quently with polysorbate 20 after first washing away the AMP or vice versa. To control for the necessity of incubating treated con-trols for a further 90 min after the first treatment, replicate sam-ples were treated with polysorbate 20 for 90 min, followed by PBS for 90 min, with a parallel group treated with PBS for 90 min, followed by polysorbate 20 for 90 min. A similar approach was taken with the magainin-2B controls. As expected, both treatment with PBS followed by magainin-2B (P⫽0.123) and magainin-2B followed by PBS (P⫽0.009) produced a slight effect, although only in the latter case was this significant. Polysorbate 20 alone produced an unexpected significant reduction in virus titer when
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
the first incubation was in PBS (P⫽4.61⫻10⫺5), possibly as a
result of increasing the incubation time to 90 min and a nonspe-cific detergent-like effect on the viral membrane. This was even more apparent when the polysorbate 20 treatment preceded the
PBS treatment (P⫽5.64⫻10⫺8), and the difference between
polysorbate 20 as the first treatment and polysorbate 20 as the second treatment in these control samples was significant (P⫽
2.17⫻10⫺5) (Fig. 2). This is likely to reflect a time-dependent
degradation of virus membranes by the detergent and fits with previous observations demonstrating the time-dependent effect of polysorbate 20 on the density of VACV virions (16).
Where samples were sequentially treated with magainin-2B and polysorbate 20, or vice versa, treatment with detergent first gave a level of virus inactivation that was similar to treatment with detergent first and then subsequently with PBS. Treatment with magainin-2B first yielded a level of inactivation that was similar to that seen when treating with PBS first and then subsequently with detergent. This indicates that treatment with AMP and detergent sequentially is equivalent to omitting the AMP. Importantly, treatment with magainin-2B and polysorbate 20 in combination, followed by a subsequent treatment with PBS, yielded a level of inactivation that was significantly greater than all other combina-tions (Pvalues ranged from 4.32⫻10⫺4to 8.67⫻10⫺8) (Fig. 2).
This suggests that magainin-2B and polysorbate 20 may act syn-ergistically to inactivate VACV and that the effect on VACV of either in isolation does not render the virus particles more suscep-tible to the other in isolation.
The observation of virucidal activity for polysorbate 20 after a 90-min incubation, while a 60-min incubation did not reduce virus titers in previous experiments (data not shown), suggested that the effects of both detergent and AMP might be time depen-dent. Accordingly, we examined the effects of polysorbate 20 and magainin-2B or LL37 separately and in combination over a range of incubation times. Magainin-2B was time dependent in its effect, taking more than 60 min to inactivate a significant proportion of the virus in the sample relative to the PBS controls (P⫽0.003) (Fig. 3A). The effect on LL37 was not time dependent in these experiments, which may reflect the greater virucidal efficacy of this AMP for VACV (Fig. 3B). The time dependency of polysor-bate 20 virucidal activity varied considerably between experi-ments. However, in all experiments when detergent and either AMP were combined, an enhanced effect was seen after a 30- or 60-min incubation, relative to PBS, polysorbate 20 alone, or AMP
FIG 1Treatment of VACV preparations with LL37 or magainin-2B. In sepa-rate experiments virus preparations in equal volumes were adjusted, to 0 () or 100 (o)g ml⫺1LL37 or magainin-2B in the absence (A) or presence (B) of
0.5% (vol/vol) polysorbate 20 for 60 min. Virus was quantitated by Reed-Muench limiting dilution. Data are presented as means and standard devia-tions of triplicate samples. Each panel is one representative ofⱖ2 experiments.
FIG 2Treatment of VACV preparations with magainin-2B in the presence and absence of polysorbate 20. Virus preparations were divided into 24 equal aliquots and treated sequentially with PBS, followed by PBS; PBS, followed by magainin-2B; magainin-2B, followed by PBS; PBS, followed by polysorbate 20; polysorbate 20, followed by PBS; magainin-2B, followed by polysorbate 20; polysorbate 20, followed by magainin-2B; or magainin-2B and polysorbate 20 together, followed by PBS (groups 1 to 8, respectively, in the figure). Magainin-2B was used at 100g ml⫺1, and polysorbate 20 was used at 0.5% (vol/vol). Sequential incubations
were for 90 min each, with the virus being washed twice in between the two treatments. After the second wash, the virus was resuspended directly in the second treatment in 50-l volumes. After the second treatment, the samples were adjusted to 200l with tissue culture medium and subjected to manual Potter-Elvejhem homogenization and titration without additional washes. Data are presented as means⫾the standard deviations of triplicate samples. The results of one representative experiment of two experiments are shown.
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
alone (Pvalues ranged from 1.1⫻10⫺5to 2.4⫻10⫺12), providing
further evidence of a synergistic interaction between the two. Synergism between AMP and polysorbate 20 suggests that the effect is mediated by alteration of the AMP secondary structure in solution, prior to AMP interaction with the target membrane. This has implications for the future development of AMPs as po-tential anti-infective agents, since it demonstrates that formula-tion in aqueous soluformula-tion can significantly affect their activity.
Polysorbate 20 increases the␣-helical content of AMPs in
aqueous solution.The apparent synergy between AMPs and
poly-sorbate 20 for inactivation of VACV led us to examine the second-ary structure of the AMPs in the presence or absence of polysor-bate 20 by CD spectroscopy. In salt-containing buffer, the magainin-2B CD spectrum shows the characteristics of an unor-dered structure with a pronounced minimum at⬃200 nm and a shoulder of negative ellipticity at⬃227 nm (Fig. 4A). These char-acteristics are also observed at 2°C, at which the magnitude of the
FIG 3Time course of AMP and polysorbate 20 activity against VACV. Virus preparations were divided into equal aliquots and adjusted to 0 or 100g ml⫺1
magainin-2B (A) or LL37 (B) and/or 0 or 0.5% (vol/vol) polysorbate 20 and incubated at 37°C for 30 (), 60 (o), 90 ( ), or 120 (p) min. Virus was quantitated by Reed-Muench limiting dilution. Data are presented as means⫾ the standard deviations of triplicate samples. The results of one representative experiment of two experiments are shown.
FIG 4CD spectroscopy. Far-UV CD spectra were recorded for magainin-2B (A) and LL37 (B) in the absence or presence of 0.40% polysorbate 20 (solid lines) at 37°C. Dashed lines represent spectra recorded at detergent concentra-tions (from the top to the bottom at 220 nm) of 0.01, 0.02, 0.05, 0.10, and 0.20% (vol/vol). Insets show the degree of␣-helicity derived from the [⌰]222
plotted as a function of the molar detergent/peptide ratios.
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
signal is about 1.5 times greater (data not shown). In the presence of 0.4% (vol/vol) polysorbate 20, the spectrum shows minima at 208 and 222 nm and a positive maximum at 192 nm, which are the characteristic CD bands for an␣-helical conformation. Spectra recorded upon titration of polysorbate 20 (from 0.01 to 0.40%) exhibit a single isodichroic point at 203 nm, which is compatible with the assumption that the coil-to-helix conversion occurs as a two-state process. Data fitting results in an apparentKd⫽0.48⫾
0.09 mM with a maximum␣-helix content of 16.9⫾0.7%. Half-maximal change in helicity is found at a 4.1:1 detergent/peptide ratio corresponding to⬃0.05% polysorbate 20 (Fig. 4A, inset).
The CD spectrum of LL-37 in salt-containing buffer is indica-tive for a pronounced␣-helical conformation (Fig. 4B), whereas in the absence of salt it shows the characteristics of an unordered structure (23). The calculated␣-helix content of 35% at 37°C is only slightly lower than the 38% observed at 2°C. The spectra are in quantitative agreement with those reported earlier (23). Upon addition of polysorbate 20, the CD intensities increase, reaching half-saturation at a detergent/LL37 ratio of⬃4.1⫾0.5 corre-sponding to aKd⫽0.23⫾0.03 mM and a maximal␣-helix
con-tent of 45.5%⫾0.3% (Fig. 4B, inset). Also in this case, spectra recorded at increasing polysorbate 20 concentrations show a sin-gle isodichroic point at 203 nm.
It is already known that AMPs adopt␣-helical conformation in target membranes. Our findings suggest that, at least for VACV, it is important that␣-helical conformation is achieved before the AMP comes into contact with the membrane. It is interesting that the effect of polysorbate 20 on␣-helical content of the AMP par-allels that of elevated ionic strength buffer, which is known to inhibit bactericidal activity of AMPs. This suggests that the en-hanced activity observed against VACV might not apply to bacte-rial targets.
The bactericidal activity of AMPs is attenuated by
polysor-bate 20 or elevated ionic strength.The parallel effects of
increas-ing ionic strength and increasincreas-ing detergent concentration on AMP␣-helix content led us to examine the effect of polysorbate 20 on bactericidal activity of the AMPs. The salt-dependent inhibi-tion of AMP bactericidal activity shown by others was confirmed withE. colibacteria treated in the presence of increasing concen-trations of NaF (to mirror conditions used in CD spectroscopy) and NaCl, with a marked inhibition of bactericidal activity at high ionic strength (Pvalues ranged from 5.7⫻10⫺9to 3.7⫻10⫺11, except for magainin-2B in the presence of NaCl, where salt con-centration did not affect AMP activity, andP⫽0.56) (Fig. 5Aand B). The effect of NaF was also examined forS. aureus, with a similar result (Pvalues of 7.6⫻10⫺9and 2.1⫻10⫺12) (Fig. 5C). WhenE. coliexperiments were repeated with polysorbate 20 in-stead of salt, an effect opposite that seen with VACV and similar to that seen against bacteria in the presence of salt was observed (Fig. 6). The bactericidal activity of the AMPs was reduced by formu-lation with polysorbate 20 (P⫽0.03 for LL37 andP⫽9.8⫻10⫺9
for magainin-2B).
These data show correlation of the effects of polysorbate 20 and salt on the bactericidal activity of AMPs just as the effects of poly-sorbate 20 and salt on AMP secondary structure in aqueous solu-tion are correlated. The correlasolu-tions suggest a mechanistic rela-tionship between the observed secondary structure and the levels of bactericidal activity.
The virucidal activity of AMPs is enhanced by elevated ionic
strength.When virucidal activity of the AMPs was assessed in the
presence of NaF, an enhancement of activity was observed analo-gous to the situation with virucidal activity in the presence of polysorbate 20 and opposite the effect observed in the treatment of bacteria (P⫽0.03 for LL37 andP⫽0.003 for magainin-2B)
FIG 5Attenuation of AMP antibacterial activity by NaF and NaCl.E. coli(A and B) or S. aureus (C) bacteria were incubated with either LL37 or magainin-2B amide at a concentration of 100g ml⫺1for 10 min in H
2O.
Peptides and vehicle (1/10 dilution of PBS in H2O) controls were adjusted to
either 500 (), 250 (p), or 0 (o) mM NaF (B and C) or 500 (), 250 (p), or 11 (o) mM NaCl (A) before the addition of bacteria. After incubation with AMP for 10 min at room temperature, samples were serially diluted in L broth and plated on L agar plates, followed by incubation overnight at 37°C. After this, the colonies were enumerated for 50% CFU using the Reed-Muench method. †, No colonies were recovered (100% kill).
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
(Fig. 7). This further correlates the effects of polysorbate 20 and salt on virucidal activity; the differential direction of potentiation against bacteria and VACV is also correlated, strengthening the interpretation that the AMP secondary structure in aqueous solu-tion is important for membrane attack.
Potentiation of virucidal activity by elevated ionic strength was also examined for a second virus, Semliki Forest virus (SFV), of the alphavirus genus of the familyTogaviridae. However, neither LL37 nor magainin-2B had any activity against SFV either in high-or low-ionic-strength buffer (see Fig. S1 in the supplemental
ma-terial). Polysorbate 20 was not included in the SFV experiments since neither AMP had baseline activity against the virus and poly-sorbate 20 inactivates the related Venezuelan equine encephalitis virus by itself (D. Ulaeto, unpublished data). There is much less data in the literature on the effect of AMPs on viruses, suggesting a general lower susceptibility of enveloped viruses to inactivation by AMPs. Interestingly, unlike SFV, which obtains its envelope from the plasma membrane, VACV obtains its outer envelope from the Golgi apparatus. The failure of the AMPs to degrade SFV does not alter our interpretation of the data with bacteria and VACV, demonstrating opposite effects of altered secondary struc-ture on microbicidal activity of the AMPs on the two types of membrane. This is highly suggestive of multiple, distinct modes of attack residing in the same peptide.
DISCUSSION
Elevated-ionic-strength buffer is known to attenuate the bacteri-cidal activity of AMPs (6,7). We have confirmed this and demon-strated in the same system that polysorbate 20 has the same effect of attenuating activity. Our interpretation is that the modulation mediated by polysorbate 20 and salt is likely to be due to their stabilization of␣-helical content in solution.
The VACV envelope is derived from the hosttrans-Golgi net-work and is thus zwitterionic, unlike the bacterial membranes of
E. coliandS. aureus, which are anionic. This would be expected to influence the interaction with AMPs, which generally carry a net positive charge (LL37,⫹6; magainin-2B,⫹4) (4). Small linear AMPs are often largely unstructured in water, although many will adopt some␣-helical structure in ionic buffer solutions and, in some cases, those with a greater propensity to form an␣-helix in solution are able to more aggressively attack eukaryotic mem-branes (8,9). To bind and insert into microbial membranes, un-folded AMPs like the magainins are thought to interact mainly via electrostatic and/or hydrophobic interactions with the lipid bi-layer. Insertion of the nonpolar side chains into the hydrophobic core of the membrane is accompanied by a structural conversion to a helical conformation and depolarization of the membrane
FIG 6Attenuation of AMP antibacterial activity by polysorbate 20.E. coli bacteria were incubated with either LL37 (A) or magainin-2B amide (B) at a concentration of 100g ml⫺1for 10 min in L broth. Peptides (
o) and vehicle (1/10 dilution of PBS in H2O) controls () were adjusted to either 0.5%
polysorbate 20 or an equal amount of diluent (H2O) before the addition of
bacteria. After incubation with antimicrobial peptide for 10 min at room tem-perature, the samples were serially diluted in L broth and plated on L agar plates, followed by incubation overnight at 37°C. After this, the colonies were enumerated for 50% CFU using the Reed-Muench method. †, No colonies were recovered (100% kill). The vehicle controls were PBS, added to a final dilution of 1:10 (vol/vol). The results of one representative experiment of two experiments are shown.
FIG 7Enhancement of AMP virucidal activity by NaF. VACV preparations were divided into equal aliquots and adjusted 0 or 100g ml⫺1LL37 or
magainin-2B, and 0 mM/(), 250 mM (o), or 500 mM ( ) NaF, followed by incubation at 37°C for 60 min. Vehicle controls were a 1/10 dilution of PBS in H2O. Virus was quantitated by Reed-Muench limiting dilution. Data are
pre-sented as means and standard deviations of triplicate samples. The results of one representative experiment of two experiments are shown.
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
through the formation of pores (24,25). In contrast to magainin-2B, LL-37 has a significant degree of␣-helix in solutions that mimic physiological salt conditions (23), probably due to the am-phipathic character of residues 11 to 32. This region contains three heptad repeats of the form (abcdefg)nwith hydrophobic residues
occupying positionsaandd; thus, this region could serve as an assembly site for forming an␣-helical coiled coil (26). It is inter-esting to speculate that the different degrees of␣-helix of LL-37 and magainin-2B in physiological buffers may be responsible for the differential activity of the AMPs in physiological buffer in the absence of polysorbate 20, whereby LL37 in physiological buffer has greater activity against VACV than magainin-2B, whereas the reverse is true for activity againstE. coliandS. aureus.
Our results showing that polysorbate 20 modulates AMP ac-tivity against bacteria and VACV in opposite directions are a strong indication that AMPs act on the different targets by differ-ent mechanisms. The correlation of the effects of polysorbate 20 and ionic strength on (i) AMP secondary structure, (ii) bacteri-cidal activity, and (iii) virubacteri-cidal activity suggest that the relation-ship between secondary structure in aqueous solution and micro-bicidal activity is causal. The data further suggest that for optimal bactericidal activity the AMP should adopt unstructured
confor-mation prior to interaction with the target membrane, and that for optimal virucidal activity it should adopt␣-helical conformation prior to interaction with the target membrane (Fig. 8). Although high␣-helical content is known to be an important feature of bactericidal AMPs (8, 9, 27, 28), prefolding into an␣-helix in solution may attenuate the ability to insert into both Gram-posi-tive and Gram-negaGram-posi-tive bacterial membranes, but enhance the ability to insert into some eukaryotic membranes. In essence, the secondary structure of the AMP in solution can be seen as increas-ing or decreasincreas-ing the effective concentration of AMP for a given target membrane without altering the absolute concentration in the preparation. Previous work has shown that linear AMPs degrade VACV membranes by a carpet-based mechanism (16), whereas bactericidal activity is more commonly ascribed to pore formation (24,25). Our data indicate that a single peptide may act by more than one mode of membrane attack. It seems possible that a carpet mechanism and pore formation are differentially favored under conditions such as those we have used and, con-ceivably, are differentially effective against the different types of target membrane.
This study suggests that formulation as an aqueous solution will present difficulties for the clinical development of AMPs for
FIG 8Schematic representation of the influence of secondary structure of a linear AMP on bacterial and viral membrane interactions. (A) In a helix-promoting formulation such as polysorbate 20 or elevated ionic strength, the AMP secondary structure is biased toward adopting an␣-helix. The helical AMP readily associates with the virus membrane, causing it to fragment. The helical AMP does not readily interact with the bacterial membrane, leaving the bacterium still viable. (B) In a helix-resisting formulation, such as low-ionic-strength buffer, the AMP secondary structure is biased toward unstructured conformation. The unstructured AMP readily associates with the bacterial membrane, converting to a helical conformation in the membrane and forming depolarizing pores. The unstructured AMP does not readily interact with the virus membrane leaving the virus still viable.
on April 29, 2016 by University of East Anglia
http://aac.asm.org/
treatment of systemic infections. However, we have described a structural requirement for AMP activity prior to association with a target membrane and how preformed secondary structure can affect selectivity for target membranes. This will facilitate wider efforts to develop complex formulations to maintain desirable secondary structure after systemic administration.
ACKNOWLEDGMENTS
We have no conflicts of interest with the contents of this article. Purchase of the CD instrument was partially funded by a BBSRC grant 75/REI18433.
FUNDING INFORMATION
UK Ministry of Defence provided funding to David O. Ulaeto. UK Min-istry of Defence provided funding to Marc A. Fox. Biotechnology and Biological Sciences Research Council (BBSRC) provided funding to Kon-rad Beck under grant number 75/REI18433.
REFERENCES
1.Jenssen H, Hamill P, Hancock RE.2006. Peptide antimicrobial agents. Clin Microbiol Rev19:491–511.http://dx.doi.org/10.1128/CMR.00056-05. 2.Marion D, Zasloff M, Bax A.1988. A two-dimensional NMR study of the
antimicrobial peptide magainin 2. FEBS Lett227:21–26.http://dx.doi.org /10.1016/0014-5793(88)81405-4.
3. Jackson M, Mantsch HH, Spencer JH. 1992. Conformation of magainin-2 and related peptides in aqueous solution and membrane en-vironments probed by Fourier transform infrared spectroscopy. Bio-chemistry31:7289 –7293.http://dx.doi.org/10.1021/bi00147a012. 4.Bourbigot S, Dodd E, Horwood C, Cumby N, Fardy L, Welch W,
Ramjan Z, Sharma S, Waring A, Yeaman M.2009. Antimicrobial pep-tide RP-1 structure and interactions with anionic versus zwitterionic mi-celles. Biopolymers91:1.http://dx.doi.org/10.1002/bip.21071.
5.Wang G.2008. Structures of human host defense cathelicidin LL-37 and its smallest antimicrobial peptide KR-12 in lipid micelles. J Biol Chem 283:32637–32643.http://dx.doi.org/10.1074/jbc.M805533200. 6.Goldman MJ, Anderson GM, Stolzenberg ED, Kari UP, Zasloff M,
Wilson JM.1997. Human-defensin-1 is a salt-sensitive antibiotic in lung that is inactivated in cystic fibrosis. Cell88:553–560.http://dx.doi .org/10.1016/S0092-8674(00)81895-4.
7.Lee IH, Cho Y, Lehrer RI.1997. Effects of pH and salinity on the anti-microbial properties of clavanins. Infect Immun65:2898 –2903. 8.Morgera F, Vaccari L, Antcheva N, Scaini D, Pacor S, Tossi A.2009.
Primate cathelicidin orthologues display different structures and mem-brane interactions. Biochem J 417:727–735. http://dx.doi.org/10.1042 /BJ20081726.
9.Zelezetsky I, Pontillo A, Puzzi L, Antcheva N, Segat L, Pacor S, Crovella S, Tossi A.2006. Evolution of the primate cathelicidin: correlation be-tween structural variations and antimicrobial activity. J Biol Chem281: 19861–19871.http://dx.doi.org/10.1074/jbc.M511108200.
10. Sørensen OE, Follin P, Johnsen AH, Calafat J, Tjabringa GS, Hiemstra PS, Borregaard N.2001. Human cathelicidin, hCAP-18, is processed to the antimicrobial peptide LL-37 by extracellular cleavage with proteinase 3. Blood97:3951–3959.http://dx.doi.org/10.1182/blood.V97.12.3951. 11. Chen H-C, Brown JH, Morell JL, Huang C.1988. Synthetic magainin
analogues with improved antimicrobial activity. FEBS Lett236:462– 466. http://dx.doi.org/10.1016/0014-5793(88)80077-2.
12. Cornut I, Thiaudiere E, Dufourcq J.1993. The amphipathic helix in
cytotoxic peptides, p 173–219.InEpand RM (ed), The amphipathic helix. CRC Press, Inc, Boca Raton, FL.
13. Sarukhanyan E, Milano G, Roccatano D.2015. Cosolvent, ions, and temperature effects on the structural properties of cecropin A-magainin 2 hybrid peptide in solutions. Biopolymers103:1–14.http://dx.doi.org/10 .1002/bip.22529.
14. Urrutia R, Cruciani RA, Barker JL, Kachar B.1989. Spontaneous po-lymerization of the antibiotic peptide magainin 2. FEBS Lett247:17–21. http://dx.doi.org/10.1016/0014-5793(89)81230-X.
15. Dathe M, Wieprecht T.1999. Structural features of helical antimicrobial peptides: their potential to modulate activity on model membranes and biological cells. Biochim Biophys Acta Biomembr1462:71– 87.http://dx .doi.org/10.1016/S0005-2736(99)00201-1.
16. Dean RE, O’Brien LM, Thwaite JE, Fox MA, Atkins H, Ulaeto DO. 2010. A carpet-based mechanism for direct antimicrobial peptide activity against vaccinia virus membranes. Peptides31:1966 –1972.http://dx.doi .org/10.1016/j.peptides.2010.07.028.
17. Butcher W, Ulaeto D.2005. Contact inactivation of orthopoxviruses by household disinfectants. J Appl Microbiol99:279 –284.http://dx.doi.org /10.1111/j.1365-2672.2005.02601.x.
18. Reed LJ, Muench H.1938. A simple method of estimating fifty per cent endpoints. Am J Epidemiol27:493– 497.
19. Gudmundsson GH, Agerberth B, Odeberg J, Bergman T, Olsson B, Salcedo R.1996. The human gene FALL39 and processing of the cathe-lin precursor to the antibacterial peptide LL-37 in granulocytes. Eur J Biochem238:325–332.http://dx.doi.org/10.1111/j.1432-1033.1996 .0325z.x.
20. Scopes R.1974. Measurement of protein by spectrophotometry at 205 nm. Anal Biochem59:277–282.http://dx.doi.org/10.1016/0003-2697 (74)90034-7.
21. Chen GC, Yang JT.1977. Two-point calibration of circular dichrometer with d-10-camphorsulfonic acid. Anal Lett10:1195–1207.http://dx.doi .org/10.1080/00032717708067855.
22. Luo P, Baldwin RL.1997. Mechanism of helix induction by trifluoroetha-nol: a framework for extrapolating the helix-forming properties of pep-tides from trifluoroethanol/water mixtures back to water. Biochemistry 36:8413– 8421.http://dx.doi.org/10.1021/bi9707133.
23. Johansson J, Gudmundsson GH, Rottenberg ME,Berndt KD, Agerberth B.1998. Conformation-dependent antibacterial activity of the naturally occurring human peptide LL-37. J Biol Chem273:3718 –3724.http://dx .doi.org/10.1074/jbc.273.6.3718.
24. Wieprecht T, Apostolov O, Beyermann M, Seelig J.1999. Thermody-namics of the␣-helix-coil transition of amphipathic peptides in a mem-brane environment: implications for the peptide-memmem-brane binding equilibrium. J Mol Biol294:785–794.http://dx.doi.org/10.1006/jmbi .1999.3268.
25. Ladokhin AS, White SH. 1999. Folding of amphipathic␣-helices on membranes: energetics of helix formation by melittin. J Mol Biol285: 1363–1369.http://dx.doi.org/10.1006/jmbi.1998.2346.
26. Crick FH.1953. The packing of-helices: simple coiled-coils. Acta Crystal-logr6:689 – 697.http://dx.doi.org/10.1107/S0365110X53001964. 27. Javadpour MM, Barkley MD.1997. Self-assembly of designed
antimicro-bial peptides in solution and micelles. Biochemistry36:9540 –9549.http: //dx.doi.org/10.1021/bi961644f.
28. Oren Z, Lerman J, Gudmundsson G, Agerberth B, Shai Y. 1999. Structure and organization of the human antimicrobial peptide LL-37 in phospholipid membranes: relevance to the molecular basis for its non-cell-selective activity. Biochem J341:501–513.http://dx.doi.org/10.1042 /0264-6021:3410501.
